OpenStack:Ironic - Revision history

Antony: fix raid node with 1 remaining disk

2024-02-21T18:42:07Z

fix raid node with 1 remaining disk

Antony: /* how to do properly install a new switch driver */

2024-02-07T16:42:26Z

how to do properly install a new switch driver

Antony: /* Hacking in a new networking generic switch driver */ formatting update

2024-01-18T11:52:20Z

Hacking in a new networking generic switch driver: formatting update

Antony: big update adding how I hacked in a new custom networking generic switch driver

2024-01-18T11:50:45Z

big update adding how I hacked in a new custom networking generic switch driver

Mariusz: /* Recommended post-deployment configuration */ Add a note on using our images with a root password set

2020-04-17T15:47:56Z

Recommended post-deployment configuration: Add a note on using our images with a root password set

Mariusz: /* Troubleshooting */ Add a description for the "TFTP open timeout" problem

2020-04-15T08:56:55Z

Troubleshooting: Add a description for the "TFTP open timeout" problem

Mariusz: /* Recommended post-deployment configuration */ Add a note on journald entries available though a HTTP endpoint during provisioning

2020-04-14T22:32:43Z

Recommended post-deployment configuration: Add a note on journald entries available though a HTTP endpoint during provisioning

Mariusz: /* Recommended post-deployment configuration */ Use /24 instead of /22 for the provisioning network

2020-04-09T12:09:30Z

Recommended post-deployment configuration: Use /24 instead of /22 for the provisioning network

Mariusz: /* Recommended post-deployment configuration */ Use a more generic value for --key-name

2020-04-09T10:19:31Z

Recommended post-deployment configuration: Use a more generic value for --key-name

Mariusz: /* Recommended post-deployment configuration */ Remove security group from the instance creation command (not needed for baremetal nodes)

2020-04-09T10:17:05Z

Recommended post-deployment configuration: Remove security group from the instance creation command (not needed for baremetal nodes)

← Older revision		Revision as of 18:42, 21 February 2024
Line 823:		Line 823:

	when you update releases you should update your fork or if you haven't done a terrible job like above push to upstream		when you update releases you should update your fork or if you haven't done a terrible job like above push to upstream
		+
		+	==Make a node that was configured in RAID back into a non-RAIDed instance==
		+	<pre>
		+	openstack baremetal node manage compute2
		+	openstack baremetal node set compute2 --raid-interface no-raid
		+	openstack baremetal node set compute2 --property root_device='{"size": "<= 2000", "rotational": false}'
		+	openstack baremetal node maintenance unset compute2
		+	openstack baremetal node provide compute2
		+	</pre>

← Older revision		Revision as of 16:42, 7 February 2024
Line 784:		Line 784:

	=== how to do properly install a new switch driver ===		=== how to do properly install a new switch driver ===
−	~~coming soon!~~	+	first step make a fork of networking generic switch and push it somewhere reachable for example here:
		+
		+	https://gitlab.define-technology.com/pub/networking-generic-switch/-/compare/stable%2F2023.1...feat-add-fs-switch?from_project_id=159&straight=false
		+
		+	this is the final working driver (for a given definition of working)
		+
		+	this driver overrides the device type in netmiko to use the existing pluribus driver. This is because of two things
		+	# it exists in the standard release
		+	# error checking is not implemented
		+
		+	The latter point is most important with this driver as I don't want to spend ages writing a full netmiko driver for this switch and it's not cisco like enough to allow those to work. It either errors going into or exiting from configure mode. If we use the pluribus driver. It never needs to go into configure mode at all and just runs the commands as is without checking the output. We know they work that is good enough for this.
		+
		+	it does NOT need to be a public repo it can be a private github repo but you need to make sure you can access it using a ssh key and that your key is on the build node or in your ssh agent
		+
		+	finally we need to buid it in to the kolla containers I made a custom kolla-build-neutron.conf file
		+
		+	<pre>
		+	[DEFAULT]
		+	namespace = kolla
		+	base = ubuntu
		+	registry = maas.pxeboot.internal:5000
		+	openstack_release = 2023.1
		+
		+	[neutron-base-plugin-networking-generic-switch]
		+	type = git
		+	location = https://gitlab.define-technology.com/pub/networking-generic-switch.git
		+	reference = feat-add-fs-switch
		+	</pre>
		+
		+	that is it
		+
		+	to build JUST the neutron-server and the prerequisite beutron-base container run
		+	<pre>
		+	kolla-build --config-file ./kolla-build-neutron.conf neutron-base neutron-server
		+	</pre>
		+
		+	then push the resulting images by hand
		+
		+	when you update releases you should update your fork or if you haven't done a terrible job like above push to upstream

@@ Line 650: / Line 650: @@
 here is my updated driver for a FS switch in a file called pluribus.py in my local dir
-<nowiki>
+<pre>
 # Copyright 2022 define-technology.
+#
@@ Line 716: / Line 716: @@
          'exit'
+     )
-</nowiki>
+</pre>
 this has properly implemented MOST  but not all of the full functionality. I'm missing a error regex so it has no way to know if a command has failed and will just plow on regardless assuming all is good buuuuut who cares about that? I am also missing a save function so the debug logs will end up being full of save not implemented but again it still works so nerrrr. It is REALLY handy to have that add/remove from trunk support as this allows you to add the vlan to uplink ports and your openstack config should be modified to include controllers and hypervisors (the latter only id DVR or similar is enabled).
@@ Line 728: / Line 728: @@
 when you have found it exit out of the container and copy your file in an restart neutron_server
-<nowiki>
+<pre>
 docker exec -it neutron_server bash
 find -iname pluribus.py | grep "kolla/venv"
 /var/lib/kolla/venv/lib/python3.10/site-packages/networking_generic_switch/devices/netmiko_devices/pluribus.py
-</nowiki>
+</pre>
-<nowiki>
+<pre>
 docker cp pluribus.py neutron_server:/var/lib/kolla/venv/lib/python3.10/site-packages/networking_generic_switch/devices/netmiko_devices/
 docker restart neutron_server
-</nowiki>
+</pre>
 if you have no typos then it should not go into a restart loop and you can try it out in your multi-tenant config
 I modified my ml2_conf.ini on the controller NOT IN KOLLA CONIFG here it is in full
-<nowiki>
+<pre>
 root@control0002:~/scratch/neutron-ngs-patch# cat /etc/kolla/neutron-server/ml2_conf.ini
 [ml2]
@@ Line 779: / Line 779: @@
 ngs_save_configuration = False
 ngs_trunk_ports = 25gigaethernet 1/0/14, 25gigaethernet 1/0/15, 25gigaethernet 1/0/17, 25gigaethernet 1/0/18, 25gigaethernet 1/0/20, 25gigaethernet 1/0/21, 25gigaethernet 1/0/27, 10gigaethernet 1/0/58
-</nowiki>
+</pre>
 I then created a new vlan provider net and the vlan was added to the switch and all the expected trunk ports were updated!
 === how to do properly install a new switch driver ===
 coming soon!

@@ Line 1: / Line 1: @@
 == Reference diagram ==
 Taken from docs for Genomics England:
@@ Line 632: / Line 637: @@
 Network interface config is exactly the same as in the previous iteration of the deployment (without multi-tenancy).

@@ Line 120: / Line 120: @@
 $ openstack image create --public --container-format bare --disk-format qcow2 --file ~/CentOS-7-x86_64-GenericCloud-1907.qcow2 centos7-1907
 </nowiki>
 Create a flavour for your baremetal nodes:

← Older revision		Revision as of 08:56, 15 April 2020
Line 419:		Line 419:

	This can happen if <code>nova-compute-ironic</code> containers are not up when you add baremetal nodes to Ironic. Make sure these containers are fully functional and then run <code>nova-manage cell_v2 map_cell_and_hosts</code> from one of the <nowiki>nova-api</nowiki> container to fix the problem.		This can happen if <code>nova-compute-ironic</code> containers are not up when you add baremetal nodes to Ironic. Make sure these containers are fully functional and then run <code>nova-manage cell_v2 map_cell_and_hosts</code> from one of the <nowiki>nova-api</nowiki> container to fix the problem.
		+
		+	------------------------------------------------------------------------------
		+	<nowiki>
		+	CLIENT MADDR: XX XX XX XX XX XX GUID: 00000000 0000 0000 0000 002590F950BE
		+	CLIENT IP: 10.6.44.249 MASK: 255.255.255.0 DHCP IP: 10.6.44.101
		+	GATEWAY IP: 10.6.44.1
		+	PXE-E32: TFTP open timeout
		+	PXE-M0F: Exiting Intel Boot Agent.
		+	</nowiki>
		+	The above message seen on the remote console indicates problems with L3 connection to the TFTP server. Make sure your switch knows how to get to the IP specified in <code>my_ip</code> or <code>tftp_server</code> parameter in Ironic conductor's config (in this example TFTP server has the 192.168.88.233 IP). Also, check if you can reach the IP assigned by Neutron DHCP to this node (in this example, 10.6.44.249) from the machine on which the TFTP server runs. If the default route on this machine doesn't go though the Ironic provisioning switch, you will need to add static routes for all your provisioning networks. For example:
		+
		+	<nowiki>
		+	# ip route add 10.6.44.0/24 via 192.168.88.253
		+	</nowiki>

	=== References ===		=== References ===

@@ Line 156: / Line 156: @@
 $ openstack server create --image <image-to-provision-node-with> --flavor baremetal.small --key-name <your-keypair> --network <name-of-the-provisioning-network> <instance-name>
 </nowiki>
 === Multitenancy ===

@@ Line 135: / Line 135: @@
   <nowiki>
 $ openstack network create public304 --provider-physical-network physnet1 --provider-network-type vlan --provider-segment 304
-$ openstack subnet create --dhcp --allocation-pool start=10.6.44.101,end=10.6.47.254 --network public304 --subnet-range 10.6.44.0/22 --gateway 10.6.44.1 public304-subnet
+$ openstack subnet create --dhcp --allocation-pool start=10.6.44.101,end=10.6.44.254 --network public304 --subnet-range 10.6.44.0/24 --gateway 10.6.44.1 public304-subnet
 </nowiki>

@@ Line 154: / Line 154: @@
   <nowiki>
-$ openstack server create --image <image-to-provision-node-with> --flavor baremetal.small --security-group ping-and-ssh --key-name mykey --network <name-of-the-provisioning-network> <instance-name>
+$ openstack server create --image <image-to-provision-node-with> --flavor baremetal.small --key-name mykey --network <name-of-the-provisioning-network> <instance-name>
 </nowiki>